Please enable JavaScript.
Coggle requires JavaScript to display documents.
Machine Learning Mathematics (Cost/Lost(Min) Objective(Max) Functions…
Machine Learning Mathematics
Linear Algebra
Matrices
Elgenvectors and Elgenvalues
Derivatives Chain Rule
Jacobin Matric
Gradient
Tensors
Curse of Dimensionality
Statistics
Measures of Central Tendency
Dispersion
Relationship
Techniques
Central Limit Theorem
Optimization
Gradient Descent
Stochastic Gradient Descent
Momentum
Adagrad
Mini-batch Stochastic Gradient Descent
Regularization
L1 norm
L2 norm
Early Stopping
Dropout
Sparse regularizer on columns
Nuclear norm regularization
Mean=constrained regularization
Clustered mean=constrained regularization
Graph-based similarity
Cost/Lost(Min) Objective(Max) Functions
Maximum Likelihood Estimation (MLE)
Cross-Entropy
Logistic
Quadratic
0-1 Loss
Hinge Loss
Exponential
Hellinger Distance
Kullback-geibier Divergence
Itakura-Saito distance
Probability
Concepts
Frequentist vs Bayesian Probability
Frequentist
Bayesian
Random Variable
Independence
Conditionality
Bayes Theorem (rule, law)
Marginalization
Law of Total Probability
Chain Rule
Distributions
Definition
Type (Density Function)
Cumulative Distribution Function (CDF)
Information Theory
Entropy
Cross Entropy
Joint Entropy
Conditional Entropy
Mutual Information
Kullback-Leibler Divergence
Density Estimation
Postly Non-parametric. Parametric makes assumpsions on my data/ranbome variables fro instance that hey are normally distributed. Non-parametric doesnot.
The methods are generally intended for description rather than formal inference.
Methods
Kernel Density Estimation
non-negative
it's a type of PDF that it is symmetric
real-valued
integral over functionis equal to 1
non-parametric
calculates kernel distributions for every sample point, and then adds all the distributions
Uniform, Triangle, Quartic, Triweight, Gaussian, Cosine, others...
Cubic Spline